EP0992903B1 - Method for implementing cooperative multitasking in a message transmission network and network element therefor - Google Patents

Abstract

A counter (CNT) and a threshold (THD) is assigned to each task (TA to TE). At the start of each processing cycle, a check is made to determine if at least one of the counters has exceeded its threshold. If it has, a task (TD) is selected and processed and its counter is reset. Those counters of the remaining tasks are increased by one count. The processing cycle is the re-run if it is determined that at least one other task (TC) is to be processed. A network element for a digital communication network is also claimed.

Description

In the field of digital communications networks, in particular in Range of SDH Networks (SDH: Synchronous Digital Hierarchy), is for each Network element contains a database in which data about the current Network configuration are stored. To manage a network element need several organizational tasks, called "tasks", in parallel be processed. These are operating systems, such as UNIX or MACH used to process the organization tasks Control access to one or more processors, with so-called "threads" are used. These operating systems are known per se and numerous, e.g. in A. Silberschatz et al. written textbook "Operating System Concepts", third edition, Cape. 4.2.3, pages 96-97, published in 1991 by Addison-Wesley, Reading, Massachusetts, USA. With the help of such operating systems is the Access from multiple organizational tasks to a processor, so controlled that for the one organizational task involving the processor currently being claimed, it is decided when the processor for the next organizational task is released. For this type of Control is also the term "cooperative multitasking" used. The Order according to which the organizational tasks are executed determines a control module, a so-called "task scheduler".

In the article "Time-Function Scheduling: A General Approach to Controllable Resource Allocation "by L. Fong et al, IBM Research Report March 1995, pages A, 1-25, describes a method for resource allocation based on the concept dynamic priorities based on different functional classes in Depending on the time to be changed, wait for a job to run got to.

From the article "Dual Priority Scheduling", R. Davis et al., Real-Time Systems Symposium 1995, pp. 100-109, is a process coordination procedure of Tasks with soft time gates in real-time systems known whose Main function is the execution of tasks with hard time limits. There are two different types of priorities for the two types of Assign tasks. Hard tasks have a priority in either one upper or lower priority range. After execution of a hard The task is placed in the lower priority area and increases in priority a fixed time back to the upper priority range. "Soft" Tasks, on the other hand, are conducted in the medium priority area.

From US 5,437,032 is such a control module and a method for Carrying out cooperative multitasking. There will be summarized several organizational tasks to so-called "jobs". Each "job" is assigned a weight according to a priority. Of the Share in the total weight of all "jobs" corresponds to the current Activity level ("actual level of multitasking activity"). It will be a desired activity level ("desired level of multitasking activity") defined, which corresponds to the maximum number of organizational tasks, which can be processed at the same time. The discrepancy between these two levels are used for the order determine after which the "jobs" are processed. Such "jobs" whose current level of activity deviates greatly from the desired one prefers. In the method known from US 5,437,032 the case occur that individual organizational tasks therefore not processed because their "jobs" only have a very small discrepancy and always put behind those "jobs" that are high Have discrepancy.

In the article "Management of SDH network elements: an application of the Information Modeling "by M. P. Bosse et al., Published in the by the Applicant published the journal "Electric Communication", Issue No. 4, 1993, at pages 329-338 are generally a method and devices for managing network elements in digital Messaging networks described. A "task scheduler" will be there but not described.

The object of the invention is a method and apparatus for Carrying out cooperative multitasking in one To propose telecommunications network, the above-mentioned Disadvantages do not have.

The problem is solved by a method with the features Claim 1 and by a network element with the features of the independent claim 9.

Accordingly, each organizational task becomes a counter and a Threshold assigned, with the threshold being a maximum number of machining cycles indicates within which the organizational task does not have to be processed immediately, and where the counter is those Machining cycles counts, within which the organizational task is not is processed. At the beginning of each machining cycle it is checked whether at least one of the counters is the appropriate threshold exceeds. If this is the case for a single organizational task, then this organizational task is processed and its counter is reset set. If this is the case for at least two organizational tasks, then so is calculated from the set of these organizational tasks according to at least a selection task, an organizational task is selected. Then this is processed and their counter is reset. Those Counters representing the remaining organizational tasks, i. not yet completed organizational tasks, are assigned to increases a counting step. The processing cycle is repeated, if it is determined that at least one of the organizational tasks to work off.

This ensures that monitors the waiting times by means of the counter and those organizational tasks that are already more than one Wait for the time period predetermined by the threshold, preferably be processed. In addition, a selection criterion is taken into account, order the sequence then, if due to the waiting times none of the organizational tasks is to be preferred.

Advantageous embodiments of the invention are the dependent claims remove.

Accordingly, it is particularly advantageous if the selection criterion Priority criterion is, according to which is examined in which order the Organizational tasks are processed. This ensures that those who perform organizational tasks whose counts are greater than the corresponding thresholds are, first of all, the very important ones Organizational tasks and then the less important Organizational tasks are processed.

It is also advantageous that the selection criterion is a time duration that indicates how long each organizational task within a waiting loop and determines according to which organizational task as next is processed. The order of processing will be so determined by the position of the organizational tasks within a Holding pattern. This so-called "round robin" principle can be very simply be incorporated into the process.

Moreover, it is advantageous if each threshold is a number that at most one less than the number N of all to be processed Organizational tasks, and if all thresholds by at least one be increased if another organizational task is to be processed. These measures will ensure that everyone Organizational tasks that have logged on for editing from benefit from cooperative multitasking and within as much as possible short time span are processed.

If the case occurs, the number N increases by one (N '= N + 1) because the further organizational task has logged on for processing, then it is especially advantageous if only those thresholds are around One increased to less than or equal to N-1. This will achieve that all organizational tasks within as few Machining cycles are processed.

It is also particularly advantageous if a single Organizational task is considered only if an event occurs, which requires the processing of this organizational task power. This means that special organizational tasks as long as can be postponed until expected events occur the processing of these special organizational tasks is required do. For example, system alerts or emergency alerts occur for which special organizational tasks are kept ready. These organizational tasks will be until the events of the Willingness to work suspended. In this very simple way and Way, the process is applied to a larger amount of Organizational tasks extends without these special Organizational tasks must be constantly managed and the Tax module ("task scheduler") is additionally charged.

The method is also particularly advantageous if a Organizational task only as one to be processed Organisotionsoufgabe is taken into account as soon as a specifiable Time interval has expired. This means that the organizational task is deactivated until the expiration of the time interval. The event is For example, the expiration of a predetermined maintenance interval and the Organizational task does a maintenance routine.

Fig. 1,

die schematisch die Struktur eines Netzelementes mit einem Steuermodul zeigt; und

Fig. 2,

die das Ablaufdiagramm eines Multitasking-Verfahrens zeigt, das das Steuermodul steuert.

In the following, the invention will be described by means of an embodiment and with the aid of the following drawings:

Fig. 1,

schematically showing the structure of a network element with a control module; and

2,

showing the flowchart of a multitasking process that controls the control module.

To simplify the following description of the embodiment the organizational tasks are also called "tasks" or "tasks" designated.

In Figure 1, the structure of a network element NE for a digital Message transmission network shown schematically. The network element NE includes a processor CPU and a control module SDR for execution of cooperative multitasking, in which the processor performs tasks according to executes an order that determines the control module. To do this the control module to a memory, not shown, in which a list is stored. The list is shown schematically in FIG. In her Only those tasks are entered which are for processing have registered. Individual tasks can be temporary or be deactivated due to the event.

As can be seen in FIG. 1, in the list for the tasks TA, TB, TC, TD and TE each entered three different parameters. The first Parameter PRY indicates with which priority the respective task to be processed. The second parameter CNT is a counter that uses each cycle of the processing cycle is increased by one. For each of the The counter therefore indicates tasks, how many processing cycles already were passed through without this task could be completed. For example, task TB already waits two processing cycles (CNT = 2) to be processed. The third parameter THD is on Threshold that will dictate and that will have a maximum number of Specifies processing cycles within which the task is not completed needs to be. For example, task TD became a threshold assigned by THD = 5. That means it can take 5 processing cycles be serviced until the task TD should be processed.

The tasks TA to TD are active, i. logged in for editing. The task TE however, it is inactive, i. not logged in and from editing suspended. The status "active" or "inactive" will pass through the list a flag SP is displayed.

As can be seen from FIG. 1, the task TA became the priority PRY = 100 and the threshold THD = 4 assigned. The meter has the current counter reading CNT = 0. This means that the task TA still wait five processing cycles until the counter CNT the Threshold exceeds THD. Only then is it indicated that the task TA should be processed as quickly as possible. The order in which the Tasks are then processed, then depends on the amount of Priority PRY. For the task TB, the following entries are added to the list find: PRY = 70, CNT = 2 and THD = 4. The entries for the task TC are the following: PRY = 50, CNT = 6 and THD = 5. For the task TD one finds following entries in the list: PRY = 60, CNT = 7 and THD = 5. And for the Task TE the parameters PRY = 30, CNT = 0 and THD = 5 are entered. This means that in the tasks TA and TB, the counters CNT the corresponding threshold THD have not yet exceeded. Of the Counter for the task TE is also below the threshold and has the Counter reading "0". For this task TE, the counter remains at zero as long as this task is disabled. The fact that the task TE is deactivated indicates the flag SP. The counters of tasks TC and TD have the corresponding Threshold already exceeded and need to be as fast as possible be processed.

Because the task TD has a higher priority PRY than the task TC TD is processed first. At the end of the machining cycle is the counter of the task TD is set to zero and the counters of the other tasks TA, TB, and TC increased by one. However, the counter remains for the task TE unchanged on the state zero until this task TE activated by an event.

The specified in Figure 1 tasks TA to TE have to edit logged in and are performed by the network element NE to To control transmissions in an SDH communication network or monitor. Each task uses one or more so-called "management objects" according to a procedure such as is known per se from the article "Management of SDH network elements: An Application of Information Modulation "by M. P. Bosse et al. published in the periodical published by the applicant "Electric Nachrichtenwesen ", Issue No. 4, 1993, pages 329 to 338.

FIG. 1 also shows a task TF corresponding to that shown there List should be added. For this task TF are the parameters Assigned to PRY, CNT and THD. With such an extension of the list the thresholds THD will be counted to the number N of all Tasks adapted. It should preferably be ensured that each threshold is at most one less than the number of all tasks to be processed (THD ≥ N-1). In the here described For example, the number of all task TA to TE equals 5 (N = 5) and becomes at Adding task TE increased to 6 (N '= 6).

The method performed by the network element shown in FIG will be described in more detail below with reference to Figure 2:

FIG. 2 schematically shows a flow chart of a method 100 to carry out cooperative multitasking. The method comprises Steps 105 to 160. In the first step 105 is a preselection to the effect that only those tasks are considered, who have registered, i. currently managed by the network element become. In this example, it is all the tasks shown in FIG TA to TE. In a next step 110, the counter is added to each task CNT and threshold THD set. In addition, each task assigned a PRY priority. These parameters are in the in Fig. 1 List shown. As already described with reference to FIG Task TA has the following parameters: PRY = 100, CNT = 0 and THD = fourth Also the other tasks TB to TE become parameters allocated according to the list shown in FIG.

Then, in a next step 115, it is queried for each task, whether its counter CNT exceeds the threshold THD. In this For example, the counter of the task TC exceeds the corresponding one Threshold. The same applies to the counter of the task TD. Through this Query will make another selection, after which in a next Step 120 only those tasks that meet this criterion, be taken into account. In this example, it is the tasks TC and TD, that need to be edited.

In order to determine a ranking, in a subsequent step 125 checked whether the priorities PRY of these two tasks TC and TD differ from each other. If this is the case, then a not shown Priority list created. In this example, the task TD is first entered in the Priority list entered because its priority PRY = 60 is greater than the Priority of task TC (PRY = 50). Thus, the task becomes TD preferably processed in a step 140.

However, if the priorities are the same, editing the tasks carried out according to the known "round robin" method. This means that the order of execution is determined by the chronological order according to which the tasks have been entered in the list. This ranking is performed in a step 130. Then the preferred task in the step 140 processed. The "round robin" method In itself, for example, in the book by H. M. Deitel "An Introduction to Operating System "described in Chapter 10.10 (second edition 1990, Publisher Addison-Wesley, Reading, Massachusetts, USA).

The one which is to be preferred to the registered tasks here first the task TD, is thus processed in step 140, where after processing whose counter CNT is reset to zero. Then In a further step 150 follows the incrementing of those counters containing the other tasks that have not yet been processed. These counters are each increased by one. Other unregistered or suspended tasks are only executed if there is no task whose Threshold is greater than zero, more occurs.

This is followed by step 160, where the proper execution the task is checked and then a return to step 115. This completes the processing cycle for a task. Then start with Step 115 is the processing cycle for the next task. This will be again checked if individual counters are larger than the corresponding ones Thresholds are. Should it be determined once in step 115 that none of the counters exceeds the corresponding threshold, then Immediately step 125 is performed, after which the differences of priorities is checked to create the priority list. Should also in this step If it is determined that all priorities are the same, then the step will be 130, after which the tasks according to the "round robin" principle be further processed.

The described method 100 ensures that all then too processing tasks are processed by the network element when individual have a very high priority. Tasks with high priorities can Do not block the editing of tasks with very low priorities. Thus, one could carry out the method described by the Designate the function of the network element as a "fair scheduler". If all Priorities are the same, according to the known "round robin "procedure is processed, thus processing the tasks the three different criteria, namely the waiting time (represents by the count CNT), the priority (represented by the Priority parameter PRY) and the classification within the Processing loop ("round robin" principle) performed.

The criterion of the waiting period takes precedence over the priority criterion and this in turn takes precedence over the "round robin" principle. It is conceivable to modify the procedure in such a way that a different ranking of the Criteria is selected. For example, the priority criterion could take precedence lifted before the criterion of the waiting period and before the "round robin" criterion. It is also conceivable to include the tests of further criteria in the Install procedure.

The method described is particularly suitable for use within an SDH network element to perform a fair scheduling. In addition, other applications are conceivable in which so far a conventional "task scheduler" was used. That is why the Invention especially for use in Unix or MACH operated Systems to be used.

Claims (10)

1. Method (100) for executing cooperative multitasking in a computer-controlled network element of a digital message transfer network, in which organisational tasks (TA to TE) are processed according to a sequence, with the following steps:

   each organisational task (TA to TE) is assigned a counter (CNT) and a threshold value (THD), the threshold value (THD) specifying a number of processing cycles within which the organisational task does not have to be processed immediately, and the counter (CNT) counting those processing cycles within which the organisational task is not processed (step 110);

   at the beginning of each processing cycle it is checked whether at least one of the counters (CNT) exceeds the relevant threshold value (THD) (step 115);

   if this is the case, then according to at least one selection criterion (PRY) one (TD) of the corresponding organisational tasks (TC, TD) is selected and processed and its counter (CNT) is reset (steps 125 to 140);

   the counters that were assigned to the remaining organisational tasks (TA, TB, TC) are incremented by one counting step (step 150);

   the processing cycle (steps 115 to 150) is run through again if it is established that at least one more of the organisational tasks (TA-TE) is to be processed (step 160).
2. Method (100) according to claim 1, in which the selection criterion is a priority criterion (PRY), according to which the order in which the organisational tasks (TA to TE) are processed is checked (step 125).
3. Method according to claim 1, in which the selection criterion is the time, which indicates how long each organisational task dwells within a waiting loop, and according to which it is determined which organisational task is next to be processed (step 130).
4. Method according to claim 1, in which each threshold value (THD) is a number that is smaller by a maximum of one than the number of all organisational tasks (TA to TE) to be processed, and in which all threshold values are incremented at least by one if a further organisational task (TF) to be processed is added (step 105).
5. Method according to claim 1, in which the number of organisational tasks is incremented by one if a further organisational task logs on for processing, and in which only those threshold values that are smaller than or equal to the increased number are then incremented by one.
6. Method according to claim 1, in which all threshold values are only incremented if an event occurs that necessitates the processing of the further organisational task (TF) (step 105).
7. Method (100) according to claim 1, in which an individual organisational task (TE) is only taken into account if an event occurs that makes the processing of this organisational task necessary.
8. Method according to claim 1, in which an organisational task is only taken into account as an organisational task to be processed as soon as a specifiable time interval has expired.
9. Network element (NE) for a digital message transfer network with at least one processor (CPU) and a control module (SDR) for executing cooperative multitasking, in which the processor processes organisational tasks (TA to TE) according to a sequence that is determined by the control module (SDR), in which to determine the sequence the control module (SDR) assigns each organisational task (TA to TE) a counter (CNT) and a threshold value (THD), the threshold value (THD) indicating a number of processing cycles within which the organisational task does not have to be processed immediately, and the counter (CNT) counting those processing cycles within which the organisational task is not processed, in which the control module (SDR) checks at the beginning of the processing cycle whether at least one of the counters (CNT) exceeds the corresponding threshold value (THD) and, if this is the case, the processor (CPU) selects and processes one (TD) from the corresponding organisational tasks (TC, TD) according to at least one selection criterion (PRY), in which the control module (SDR) resets its counter (CNT) and at the end of the operating cycle increments those counters assigned to the other organisational tasks (TA, TB, TC, TE) by one counting step, and in which the control module (SDR) controls a fresh run-through of the operating cycle if it establishes that at least one more of the organisational tasks is to be processed.
10. Network element according to claim 9, in which several processors linked to one another access several databases to process the organisational tasks according to cooperative multitasking, and in which one of the processors accesses the control module to determine the order for processing the organisational tasks.

Images